Proof Composition for Deductive Verification of Software Product Lines

TLDR: This work introduces a classification of existing software product line verification approaches, proposes proof composition as a novel approach and presents a case study to evaluate proof composition and demonstrates that it reduces the effort for verification.

Abstract: Software product line engineering aims at the efficient development of program variants that share a common set of features and that differ in other features. Product lines can be efficiently developed using feature-oriented programming. Given a feature selection and the code artifacts for each feature, program variants can be generated automatically. The quality of the program variants can be rigorously ensured by formal verification. However, verification of all program variants can be expensive and include redundant verification tasks. We introduce a classification of existing software product line verification approaches and propose proof composition as a novel approach. Proof composition generates correctness proofs of each program variant based on partial proofs of each feature. We present a case study to evaluate proof composition and demonstrate that it reduces the effort for verification.